Discussion

In the present study the aim was to develop analytical methods that would enable the pre- concentration, detection and quantification of known EDCs in the environment, as tool to evaluate the JTED Ecomachine. Three highly sensitive analytical instruments were identified to detect and quantify the pre-concentrated EDCs from the environment, namely a UPLC-, UPC2- and GC-MS/MS system. Methods were developed on each instrument and optimized by changing variables such injection solvent, column temperature, eluent flow rate, and desolvation gas temperature and flow rate, before selecting the optimal conditions. The optimized instrumental methods were then evaluated for their sensitivity.

Together with their feasibility, i.e. how many of the selected EDCs can be detected and quantified, and economic friendliness one instrument was selected for the final method validation. Next, a SPE protocol was developed that would enable the pre-concentration of the compounds of interest from environmental samples. Here, two SPE cartridges with different stationary phase chemistries were considered. In addition, an optimum extraction pH (2, 4, or 6) was examined. Second to last, a preliminary recovery study for the pre- concentration protocol was performed to determine its feasibility before extensive method validation. Finally, an extensive method validation study was carried out to evaluate the pre-concentration, identification and quantification protocol as whole.

Method development was initiated by first setting up the relevant SRMs or MRMs on each instrument. This required the direct injection of each derivatized compound on each instrument separately. Table 3.2 shows the optimized MRMs for each compound set up on both the UPLC- and UPC2-MS/MS. For UPLC- and UPC2-MS/MS the base peak observed for all compounds except CBZ-d10 and TCC was that op the product ion at m/z 171. The main secondary product ion for these compounds was found to be at m/z 156, with the exception of BPA whose secondary ion was selected to be at m/z 170. These product ions corresponds to previous studies utilizing DNCl as derivatizing agent (41, 358, 359). The product ion for BPA has not been reported before using an ESI mass spectrometer. TCC and CBZ-d10 did not show these product ions as these compounds do not get derivatized with DNCl under the conditions prescribed. The product ions m/z 160 (base peak) and m/z 126 observed for TCC have been report by Hancock et al. (360) and Jongmun et al. (361). The product ions found for CBZ-d10 corresponded to an m/z at 202 and 204 (base peak), which is 10[2H] heavier than the reported m/z 192 and m/z 194 reported for CBZ (362, 363). Finally, each compound was injected on the relevant columns and the retention time on each column noted in Table 3.1. From the results in Table 3.1 it is evident that sufficient separation was achieved for all compounds on all UPC2 columns. However, TCC eluted at nearly the same retention time as that of E3 and MeP on the UPLC column. Both E3 and MeP MRM scans were performed in positive mode and therefore greatly influenced the sensitivity of the method for TCC. This is evident from Table 3.7 showing that TCC could not be detected when a preliminary sensitivity study was performed. For GC-MS/MS analysis Table 3.3 shows the relevant precursor and product ions as well as each compound‘s retention time on the column. The different precursors or product ion were matched to that found in literature (364–371). Like the UPC2-MS/MS, sufficient separation of compounds could also be achieved on the GC- MS/MS.

After method development, certain variables were investigated to improve the peak area, peak shape and S/N. The optimized conditions are laid out in Table 3.4, Table 3.5 and Table 3.6. According to Kruve et al.(43), of the parameters being investigated on the UPLC, injection solvent composition/derivatizing solvent have the highest likelihood to have an effect on parameters such as peak shape, area and S/N. However, the current study showed that column temperature played a bigger role (Error! Reference source ot found., Error! Reference source not found. and Error! Reference source not found.). Nevertheless, injection solvent composition significantly affected the peak area of all compounds investigated. This was also the case for desolvation gas temperature. Next to column temperature, desolvation flow rate had the greatest effect on peak shape, while all variables contributed similarly to the effect on S/N. A possible explanation for the larger role column temperature played in results could be explained by the temperature at which derivatization is normally carried out. The optimal heat conditions resembling that of the derivatization conditions could help influent compounds to be at the correct temperature for all to be at their derivatized masses before injection into the mass spectrometer. On the UPC2-MS/MS, injection solvent composition affected the parameters the least, while all other variables played an equally important role on parameter improvements/impairments (Error! Reference source not found., Error! Reference source not found. and Error! erence source not found.). The smaller effect injection solvent composition played suggests that the UPC2 instrument may be more compatible with a diverse range of injection solvents. Lastly, flow rate showed to be the most effective in improving the overall performance of the method. In contrast to both UPLC- and UPC2-MS/MS analysis, injections solvent composition and split mode on GC-MS/MS displayed the greatest effect on parameters (Error! Reference source not found., Error! Reference source not und. and Error! Reference source not found.). Although it is impossible to pinpoint the main contributor to parameter improvements between solvent composition and split mode without testing the conditions separately, pulsed splitless injections have been shown to improve sample transfer and enable trace level analysis (372–375).

Before samples could be analysed using either developed instrumental method, a preliminary sensitivity study needed to be done to determine which system – UPLC/UPC2 would yield the best results in future. This include being able to optimize the EDC pre- concentration protocol. Therefore, a dilution series was prepared consisting of a mixture of all the compounds under investigations. The dilution series was injected on the UPLC- and GC-MS/MS and preliminary LODs, LOQs, ULOQs and linearity‘s for the different compounds calculated. Results are shown in Table 3.7 and Table 3.8. The obtained

results suggested that the UPLC/UPC2-MS/MS instruments have higher sensitivities than the GC-MS/MS instrument and therefore latter was excluded from further investigations and the final extensive validation. While in the process of deciding which of the two instruments – UPLC or UPC2 to use for the optimization of EDC extraction protocol from environmental matrices, the then used UPC2 2-EP column failed and no other columns were available for use. This forced the use of the UPLC-MS/MS as instrument of choice for the use of optimizing the EDC pre-concentration protocol.

During the first trials of constructing an EDC pre-concentration protocol, the Supelclean ENVI-18 – a C18 cartridge, and a Supel-select HLB – a hydrophilic and lipophilic cartridge, were under investigation. Extractions were carried out on each and results showed that the HLB cartridge could retain compounds at a higher percentage MeOH wash step than the ENVI-18 (Error! Reference source not found.). Most EDC nvestigated in this study are hydrophobic in nature. However, many of the EDCs contain hydroxyl groups and therefore will interact better with a stationary phase also containing polar groups. In the final step of constructing the EDC pre-concentration protocol the pH at which compounds would be optimally retained on the HLB cartridge was investigated. Here different organic acids were used to acidify both the equilibration solvent as well as the sample. Error! Reference source not found. shows the preliminary recoveries under ifferent sample and equilibration solvent conditions. Statistical analysis revealed no noteworthy differences between the different acidification steps. However, extensive work by Baker and Kasprzyk-Hordern suggest acidifying samples to a pH of two for pharmaceuticals and illicit drugs in surface and waste water (376). From this suggestion the studies that followed used a pH of two. Finally, a more comprehensive recovery study was performed to assess the EDC pre-concentration protocol performance. Here an additional matrix was added that represented environmental samples more closely. Table 3.9 shows the estimated recoveries to be expected during method validation. For both filtered and unfiltered water acceptable recoveries were obtained for 8 of the 12 compounds. However, under both filtration and no filtration the SWW matrix showed recoveries never achievable. This suggests that the complex matrix might have ionization enhancing effects and therefore would first need to be quantified before accurate recoveries can be determined. The acceptable recoveries obtained from water allowed the proceeding to extensive method validation.

Finally, the pre-concentration protocol needed to be validated together with either the UPLC- or UPC2-MS/MS instrument. While evaluating the data from the preliminary

recovery study a new column for the UPC arrived. Although the column chemistry differed, it was decided to do the final extensive method validation on the UPC2-MS/MS for the following reasons: First and foremost, this is ‗n newly available instrument that will need to be reviewed. Therefore, completing the validation on this instrument will serve as publicity. Second, between the UPLC and the UPC2, only the UPC2 system was able successfully separate all compounds. Second to last, the UPC2 has a reduced solvent consumption and therefore are a cheaper alternative to the UPLC. Finally, as this thesis is linked to environmental studies and the UPC2 provides greener chemistry, this instrument will fit well into being environmentally friendly.

Extensive validation of the EDC pre-concentration, identification and quantifications was performed by investigating the conditions of sensitivity, accuracy, precision and trueness as set out in guidelines by Kruve et al. (43, 44). Using the sensitivity data from the preliminary studies as a guideline, calibration curves were generated to determine the LODs, LOQs, ULOQs and linearity‘s of each compound. Table 3.10 shows the relevant information. From the table it can be gathered that the sensitivity data obtained from the method validation correlate with that of preliminary findings, with only MeP showing deviations from earlier findings. Furthermore, accuracy data confers the ability of the instrument to measure all compounds qualitatively within acceptable limits (Table 3.10). In addition, each measurement was precise for all matrices tested and the deviations measured neglectable (Table 3.10). Finally the stability data in Table 3.10 show that overall all compounds, with the acception E1, E3 and and TCC, are stable for at least 17 hours. The latter confer minimal change to compounds and therefore infer reliability. To quantify the possible influence that environmental samples could exert on sample preparation and the final result, the matrix effects was investigated. Results are shown in Error! Reference source not found.. From these data we find the matrix effect for most ompounds are high and that results obtained from e.g. a WWTP could vary greatly. This data is supported by earlier finding on recovery that showed recoveries for some compounds that are never achievable. Finally, the method was evaluated for sample preparation effectiveness and the robustness of the EDC pre-concentration method. Process efficiency data in Table 3.11 show high variability between the different cartridge loads and investigators. This marks the process – from extraction to injection, as complex with room improvement and simplification. This view is strengthened by the results of Table 3.12 that show good to excellent recoveries obtainable for a number of compounds, in both water and SWW matrices. These results therefore highlight the care that should be

taken in processing samples from complex matrices. Additionally, it also emphasizes that each step in the processing of samples needs to be done quantitatively.

To conclude, methods were set up on three analytical instruments and through a process of elimination the UPC2-MS/MS was selected for the final method validation. Additionally, an EDC protocol developed, optimized and establish. Finally, the method was validated and found to be highly sensitive, accurate, precise and relatively robust, but that it lacked simplicity. From this information it is evident that the developed method can be used for environmental analysis such as the testing of samples from the JTED Ecomachine. However, it is recommended that the EDC pre-concentration protocol be simplified by doing away with steps such as the transfer between vials, as well being further optimized.